* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Evaluation of screening test
Survey
Document related concepts
Fetal origins hypothesis wikipedia , lookup
Race and health wikipedia , lookup
Eradication of infectious diseases wikipedia , lookup
Audiology and hearing health professionals in developed and developing countries wikipedia , lookup
Differential diagnosis wikipedia , lookup
Preventive healthcare wikipedia , lookup
Epidemiology wikipedia , lookup
Newborn screening wikipedia , lookup
Public health genomics wikipedia , lookup
Transcript
Screening Dr. Pramod Sah Framework 1. 2. 3. 4. 5. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. Definition Objectives of screening Difference between screening and Diagnostic test Requirements of tests used for Screening Measures of the validity of a screening test: a. Sensitivity b. Specificity Relationship between sensitivity and specificity. Measures of the performance (yield) of a screening test: a) Predictive value positive (PV+) b) Predictive value negative (PV-) Factors influencing PV+ and PVBayes’ theorem ROC curve Kappa Statistics Considerations for a screening programme Evaluation of Screening Programmes Biases in Screening Programmes References One of the priority duties, not only of a public health physician but for all medical personnel is to ensure an early diagnosis and treatment, through “screening for the disease” Definition The presumptive identification of unrecognized defect or disease by application of tests, examinations or procedures which can be applied rapidly, to sort out apparently well persons who probably have a disease, from those who probably do not.(commission on chronic disease 1951) o i.e, screening is the use of tests to help diagnoses(or their precursor conditions) in an earlier phase of their natural history or at the less severe end of the spectrum than is achieved in routine clinical practice. o It attempts to uncover the iceberg of disease. A screening test is not indented to be a diagnostic test. It is only an initial examination. Persons with positive or suspicious finding must be referred to physicians for diagnosis and necessary treatment. 1 Why Screening is important : 1) Diagnostic and therapeutic advances are often slow, but screening may be a “direct solution” to modify history of a disease in a population. 2) It provides a model for studying disease mechanisms and the natural history of a disease. What is screening program and screening test: 1. Screening program: Comprehensive disease control activity based on the identification and treatment of persons with either unrecognised disease or unrecognised risk factors for disease. 2. Screening test: specific technology (survey questionnaire, physical observation or measurement, laboratory test, radiological procedure) used to help identify persons with unrecognized disease or unrecognized risk factors for disease. The distinction between screening, case-finding or diagnosis should be clear-cut. Often however, it is blurred by multiplicity of tests used and the haphazard nature of diagnostic decision-making. Screening Screening is testing for infection or disease in population or in a individuals who are not seeking health care Ex- serological testing for AIDS virus in blood donors, neonatal screening Case-finding Is use of clinical and /or laboratory tests to detect disease in individuals seeking health care for other reasons. Ex- VDRL test to detect syphilis in pregnant women, pulmonary tuberculosis in chest symptomatic. Difference between screening and diagnostic test: Screening test Healthy clients Applied in group Criterion fixed Less Accurate Inexpensive Not a basis for treatment Initiative from provider Diagnostic test Sick clients/ indication Applied to single patients Criterion flexible More Accurate More Expensive Basis for treatment Initiative from patients 2 Objective of screening: The basic purpose of screening is to sort out from large group of apparently healthy persons those likely to have the disease or at increased risk of the disease under study. (To lower morbidity and mortality of the disease in a population (control, rather than elimination of disease). Screening in health care is undertaken with the following broad objective; 1. To ensure early detection of a disease among individuals, so that prompt treatment may be instituted; e.g screening for cervical, breast cancer, hypertension etc. This is also called “PRESCRIPTIVE SCREENING” . 2. To protect the community from disease that the persons being screened has, also called “PROSPECTIVE SCREENING”. Screening the blood units for HIV. 3. For entry into certain forms of occupations (armed services, industries, etc) with a view to “weed out” those who are unfit or whose existing health status may be adversely affected by occupational conditions. 4. Screening is sometimes done to help allocate health care resources. 5. Screening may be done simply for research, for example, to identify disease at an early stage to help understand the natural history. . Characteristics of disease appropriate for screening: 1. Prevalence of pre-clinical disease should be relatively high among those screened 2. The disease should be serious(relates to cost effectiveness, ethics, and prognosis) 3. Natural history of disease should be understood, such that the detectable sub-clinical disease stage is known and identifiable. 4. Pre-clinical disease left untreated typically progresses to clinically-evident disease(e.g. no spontaneous regression) 5. Early detection of disease leads to more favourable prognosis due to early treatment, as compared to delayed treatment. Requirements of tests and used for screening A screening test should be: Valid: it should be “accurate” i.e. should measure correctly what it intends to. It should have high sensitivity, specificity, and positive and negative predictive values. Reliable (Precise): it should give consistent results when repeated applications are made. 3 Yield: It should give enough number of cases to commensurate with the expenditure and inputs involved. Yield will depend on sensitivity of the test, prevalence of the disease (if screening is applied to a high risk group, the yield will be better) and availability of medical care (if medical care has not been available to the community being screened, a large number of people with the disease will be diagnosed). Practical: The test should be easily administered by even persons with ordinary training, should be innocuous, acceptable and should give fairly quick results. Efficient: The amount of input (in terms of expenses and time) should result in reasonable amount of output in terms of improved health and satisfaction Types of screening 1. Mass screening 2. High risk or selective screening 3. Multiphasic screening 1. Mass screening: simply means the screening of a whole population or a sub – groups. Ex- all the school children’s. It is offered to all, irrespective of the particular risk individual may run of collecting the disease in question (e.g. - Tuberculosis) 2. High risk or selective screening: screening will be more productive if applied selectively to high risk group, defined on the basis of epidemiological research. Ex- cervical cervix tends to occur relatively less often in the upper social groups, screening for cancer cervix in the lower social groups could increase the yield of new cases. 3. Multiphasic screening : Application on two or more screening tests in combination to a large number of people at one time than to carry out separate screening tests for single disease. Uses of multiple tests Serial (sequential) screening test : Two screening tests can be applied in serial one after the other, by taking people who tests positive on the first test for the second test. In fact the same test (e.g. ELISA for HIV) can also be done two times in serial. This procedure will greatly increase the specificity and positive predictive value, but lead to decline in net sensitivity. Parallel (simultaneous) screening test: two tests can be applied in parallel and the person can be considered as positive if any one or both of the tests are positive which leads to an increase in net sensitivity but a decline in net specificity. 4 Evaluation of screening test: 2x2 table By convention, the real status, gold standard or reference test is at the top, while the diagnostic test being assessed is down the side. Two by two table for screening tests Test Disease Present Absent a (True positive) b (False positive) c (False negative) d (True negative) a+c b+d + - Total Total a+b c+d N= a+b+c+d Sensitivity = Ability of the test to call positive those who really have the disease. or True positive rate or Positivity in disease. = No of true positives that are detected / Total no with disease = Chance of +ve test, given +ve disease = a/ (a+c) Answers the question “how good is this test at picking up people who have the condition”. Specificity = Ability of test to call negative those who do not have of disease or True negative rate or Negativity in health = No of true negatives that are detected/ total no with disease = Chance of negative test, given no disease = No of true +ve/ No of people without the disease = d/ (b+d) Answers the question “how good is this test at correctly excluding people without the condition”. Ideally a test should be both highly specific as well as highly sensitive, so as to get the maximum numbers of True Positives (TP) and True negatives (TN). However, in practice this is not possible and some trade off, depending on the disease under consideration is required. If a disease has serious implications if not treated early and has adequate treatment available, then our aim would be to not leave out any “false negative”, even at the expense of getting many false positives; in such instance we would go up for highly sensitive test. 5 The mnemonics SnOut and SpIn provide some guidelines on how to interpret sensitivity and specificity for an individual patient. SnOut helps physicians to remember that a highly S ensitive test with a n egative result is good at ruling-out the disease. SpIn reminds physicians that a highly S pecific test with a p ositive result is good at rulingin the disease. RELATIONSHIP BETWEEN SENSITIVITY & SPECIFICITY: 1. Lowering the criterion of positivity results in increased sensitivity, but at the expense of decreased specificity. 2. Making the criterion of positivity more stringent increases the specificity, but at the expense of decreased sensitivity. 3. The goal is to have both high sensitivity and high specificity, but this is often not possible or feasible. 4. In general, specificity must be at least 98%to be effective -- because misclassifying 2% of the population will create as many false positives as the sensitivity of the test will actually detect. 5. Sensitivity should be increased when the penalty associated with missing a case is high (e.g. minimize false negatives) --- when the disease can be spread --- when subsequent diagnostic evaluations are associated with minimal cost and risk 6. Specificity should be increased when the costs or risks associated with further diagnostic techniques are substantial (minimize false positives – e.g. positive screen requires that a biopsy be performed). The problems with sensitivity and specificity. Suppose if one of our patients has tested positive on ELISA, what are his chances of really having HIV infection? OR if he has tested negative, what are the chances that he really does not have the HIV infection? These questions clearly cannot be answered by sensitivity and specificity. In this situation we use the predictive value. Predictive value- Sensitivity and specificity of any diagnostic test are fixed, i.e. they will not change. However, our interest in screening is not only the sensitivity or specificity but rather the predictive values; i.e. if an individual has tested positive on a screening test, what are the chances that he really does have the disease. This is called the Positive Predictive Value (PPV or PV+). Unfortunately, predictive values are highly dependent on the prevalence of that disease in the population being screened. The same test with the same levels of sensitivity and specificity will give a very high PPV if the prevalence of disease is high but a very low PPV if prevalence is low. The optimum prevalence to get a very high PPV as well as NPV is between 30 - 60%. Thus, the health administrator should aim at getting the population to be screened in such a way that prevalence of the condition for which screening is being undertaken, is between 30% to 60%. 6 Whenever we use any “diagnostic test” (including clinical signs), at the very outset we have to accept the fact that we will never get 100% accurate results. Some deviation on either side (i.e. the test is positive but disease is really absent or the test is negative but the disease is really present), is always likely to occur. Positive predictive value = chances of having the disease, given that the test is +ve (post test probability of a positive test). = No with +ve test and disease / all people with +ve test = a / (a + b) Answers the question “if a person tests positive, what is the probability that he or she has the condition?” Negative predictive value = chances of not having the disease, given that the test is –ve (post test probability of a positive test). = No with ve test and no disease / all people with + ve test = d / (c + d) Answers the question “if a person tests positive, what is the probability that he or she has the condition?” Accuracy = True positives and true negatives of a test as a proportion of all results. Answers the question “what proportion of all tests have given the correct result?” Bayes' Theorem and Likelihood Ratios Bayes' Theorem Adapting a theory of conditional probability from the 18th century statistician Thomas Bayes solves the problem of calculating post-test disease probability. This theory allows pre-test probability to be separated from a term that describes the strength of the diagnostic test— likelihood ratio. Bayes' theorem: Post-test Odds = Pre-test Odds X Likelihood Ratio 7 Likelihood Ratio Likelihood Ratio is a ratio of two probabilities Likelihood ratio is equal to the probability of the observed test result among the diseased divided by the probability of the same test result among the non-diseased Likelihood ratios take into account both the pre-test probability of a disease (this is reflected by the baseline prevalence of the disease) and the post- test probability (this is reflected by the positive and negative predictive values of the test). Likelihood ratio of a positive result = Sensitivity/(1 – specificity) Answers the question “How much more likely is a positive test to be found in a person with the condition than in a person without it?” (Likelihood ratios are now considered the most important findings). If prior probability of disease known for an individual patient, then posterior probability of disease can be calculated using a nomogram developed by Sackett and colleagues. Test may not prove presence or absence of disease, but can give more accurate probability of presence or absence of disease. Likelihood ratio of a negative result = (1 – sensitivity)/specificity Answers the question “How much more likely is a negative test to be found in a person without the condition than in a person with it?” (Likelihood ratios are now considered the USE OF LIKELIHOOD RATIO Employment of following three step procedure: 1. Identify and convert the pre-test probability to pre-test odds. 2. Determine the post-test odds using the formula, Post-test Odds = Pre-test Odds * Likelihood Ratio 3. Convert the post-test odds into post-test probability. Since likelihood ratios are calculated from sensitivity and specificity, LRs are stable operating test characteristics, unaffected by prevalence of disease. A LR of 1.0 is a useless test because this result fails to change the opinion of probability of disease from pre-test to post-test. LR+ are always greater than 1.0; the larger the number, the more likely is the patient to have the disease after a positive test result. LR- are always less 8 than 1.0, with the smaller numbers signifying a lower risk for disease than pre-test estimates.. Strength of the Test by Likelihood Ratio Qualitative Strength LR(+) LR(-) Excellent 10 0.1 Very good 6 0.2 Fair 2 0.5 Useless 1 1 Using Bayes' Theorem Bayes' theorem: Post-test Odds = Pre-test Odds X Likelihood Ratio This form of Bayes’ theorem using likelihood ratios requires the conversion of pre-test probability to odds multiplied by the appropriate LR and then reconverted to the post-test odds, back into post-test probability. Example: What is the probability of a pulmonary embolism in a patient after a positive CTA VN (sensitivity 90%, specificity 95%) if the patient has a pre-test probability of 28%? Bayes' theorem: Post-test Odds = Pre-test Odds X Likelihood Ratio Steps Calculations 1. Convert pre-test probability to odds. Pre-test odds = 0.28 / (1 - 0.28) = 0.389 Odds = Probability / (1 – Probability) 2. Calculate LR+. LR+ = 0.90 / (1 - 0.95) = 18.0 LR+ = Sensitivity / (1 – Specificity) 3. Calculate Bayes' Theorem. Post-test odds = 0.389 X 18 = 7.0 Post-test Odds = Pre-test Odds X LR 4. Convert post-test odds to probability. Pos-ttest probability = 7 / (1 + 7) = 87.5% Probability = Odds / (1 + Odds) 9 “Receiver Operator Characteristics (ROC) curve ROC curve is a graphic method of assessing the ability of a screening test to discriminate between healthy and diseased persons. ROCs are particularly useful in evaluating those tests which do not give a categorical positive or negative result but instead give a value. The ROC plots the different levels of sensitivity and specificity for the differing cut off points. The cut off which gives the maximum sensitivity and maximum specificity will be the best cut off to use in practice. For a screening test in which the result is measured on a continuous scale, the decision regarding cut off point would need consideration by experienced clinicians and public health administrators. For example, if tonometry is being used as a screening test for glaucoma, the results (18, 18.5, 22, 26.8 mm etc.) would be measured on a continuous scale. Now, possibly most of the eyes with intra ocular pressure upto 22 may be, in fact, non - glaucomatous while most of the eyes with intra ocular pressure more then 26 mm will be, in fact, really glaucomatous. The main problem will be the eyes having ocular pressure between 22 and 26 mm Hg, i.e. the area of overlap between glaucomatous and non glaucomatous eyes. If we keep the cut - off point low, say at 22 mm Hg, we will correctly identify nearly all the glaucomatous eyes (high TP) but will also identify a number of normal eyes as glaucomatous (large numbers of False Positives (FP)). On the other hand, if we keep the cut off at 26 mm Hg, we would correctly identify nearly all the normal eyes (high TN) but would miss out a large number of glaucomatous eyes who have mild rise of pressure (large numbers of False Negatives (FN). In the former instance, when we have a low cut off point (i.e. less stringent criteria) we have, thus, higher sensitivity but lower specificity. On the other hand, in the latter instance, when we have a higher cut - off point (i.e. more strict criteria) we get a higher specificity at the cost of a lower sensitivity. In practice, the public health decision regarding the level of cut off is done based on the potential for treatment following early detection, the actual availability of treatment and the “labeling effect of having a disease”, if any. In such cases, advanced statistical techniques in the form of “Receiver Operator Characteristics (ROC) curve analysis” are available to scientifically work out the optimum cut off point, which gives the best trade off between specificity and false - Positives (i.e. 1 - Specificity). 10 Receiver Operating Characteristic (ROC) curve -- sensitivity against 1- Specificity 11 Cut-points and Receiver Operating Characteristic (ROC) in Continuous Data Kappa Statistics • • • • The choice of the methods of comparison is influenced by the existence and/or practical applicability of a reference standard (golden standard). If a reference standard (golden standard) is available, we can estimate sensitivity and specificity – ROC (receiver operation characteristics) analysis. If a reference standard is not available or there is no golden standard for comparison, we cannot perform ROC analysis. Instead, we can assess the agreement and calculate the Kappa. For example, for comparing the assessment from two different readers, we would calculate Kappa, overall percent agreement, positive percent agreement, and negative percent agreement Kappa Statistic(K) is a measure of agreement between two sources, which is measured on a binary scale (i.e. condition present/absent). K statistic can take values between 0 and 1. 12 How to measure Kappa . Me Measurements by observer B Observer B Positive Negative Total Measurements by observer A Positive Negative Total a c a+c a+b c+d a+b+c+d =N b d b+d • • • Proportion that A scored positive = a+c/N Proportion that B scored positive = a+b/N Therefore by chance alone it would be expected that the proportion of subjects that would be scored positive by both observers = (a+c/N. a+b/N) Proportion that A scored Negative = b+d/N • • Proportion that B scored positive = c+d/N Therefore by chance alone it would be expected that the proportion of subjects that would be scored negative by both observers = (b+d/N. c+d/N) • Therefore, total expected proportion of agreement = (a+c/N. a+b/N) +(b+d/N. c+d/N) = Pe • • Maximum proportion of agreement in excess of chance = 1- Pe Total observed proportion of agreement in excess of chance = a+d/N =Po • Therefore, proportion of agreement in excess of chance = Po – Pe • The observed agreement in excess of chance, expressed as proportion of the maximum possible agreement in excess of chance(Kappa) is = K = Po – Pe/1-Pe 13 Scales for Strength of Agreement for Kappa FLEISS < 0.4 0.4 – 0.75 >0.75 • • LANDIS & KOCH poor Fair to good excellent 0.0 - < 0.2 0.2- < 0.4 0.4 - < 0.6 Slight Fair Moderate 0.6 - < 0.8 Substantial 0.8 – 1.0 Almost perfect If the observers are in complete agreement then κ = 1. If there is no agreement among the observers other than what would be expected by chance (as defined by Pe), κ = 0. Considerations before Launching a Screening Programme Screening in public health settings should only be launched after carefully considering various aspects. Considerations for a screening programme 1. 2. 3. 4. 5. 6. 7. The condition should be an important health problem. There should be an acceptable and effective treatment. Facilities for confirming the diagnosis and for treatment should be available. There should be recognizable latent / early symptomatic stage. There should be a suitable screening test or examination available. The test should be acceptable. The natural history of the condition, including development from latent to apparent disease, should be adequately understood. 8. There should be an agreed policy regarding whom to treat as patients. 9. The cost of case finding (including final diagnosis and treatment) should be economically balanced vis - a – vis the expenditure on medical care as a whole. 10. Case finding should be a continuing process and not “once and for all” project. Detection of cancer of uterine cervix using “pap test” is a procedure which meets all these 10 criteria. The test is based on the assumptions that, firstly, a high proportion of cancer cervix detected in situ would progress to invasive cancer over time; secondly, most cancers remain in situ long enough for screening at reasonable intervals to detect a high proportion of cancer cases; and, thirdly, carcinoma in situ is highly curable. 14 Other diseases which are amenable to screening include breast cancer, Hypertension, Anemia during pregnancy, Diabetes Mellitus, growth screening in children, CHD screening in high risk groups, phenylketonuria among new born etc. Evaluation of Screening Programmes Contemporary medical evidence strongly recommends that the effectiveness and impact of screening programmes must be evaluated by Randomised Controlled Trials (RCTs) by comparing the outcome measures between the screened and unscreened groups. Biases in Screening Programmes Lead time bias: Lead time is defined as the interval between the point a condition is detected through screening and the time it would normally have been detected due to appearance and reporting of signs and symptoms. If early detection has no effect on the course of disease then it will be like giving the patient a few more years of sickness and apprehension rather than health! (e.g. HIV detection). In such cases, it is possible that screening, through earlier detection, will advance the time of diagnosis without delaying time of death, thereby increasing the “diagnosis - to – death - time” and tend to show “increased survival” among the screened group as compared to the group not given screening test, though in reality there would be no increase in survival. Length bias: It has been observed that cases detected through periodic, early detection programs, tend to have longer preclinical stages than those missed out by screening but self detected between examinations. This preclinical stage is defined as the interval between the time a screening test is capable of detecting disease and the time the patient seeks care as a result of experiencing symptom detected patients. Thus, the length bias tends to spuriously show a better survival among screen detected cases. Self selection bias: If the groups which are offered (and the other which is not offered the screening) are not constituted by random allocation but rather on the basis of self selection (volunteers), it is possible that such volunteers may be more health conscious, educated and more likely to give up associated risk factors; hence survival in such a screened group is likely to be better, not due to screening but because of associated factors. Public Health Officer’s Check List while Planning a Screening Programme In the past, many screening programmes have been launched simply due to over enthusiasm, without any consideration to the epidemiological facets of proper health planning. The result has been often quite adverse, creating unnecessary public aversion towards screening programmes (due to lack of proper diagnostic test or treatment), and wastage of resources. It is therefore necessary that the Public Health Officer in charge of Community health care should check the following list sequentially, while launching a screening programme. Do a “situational analysis”: Undertake a quick collection of information, by going through existing records of health institutions and other governmental/non governmental agencies, or 15 else, collect information by a quick survey, in respect of the community to be screened about the: 1. Demographic profile; 2. Attitudes towards utilization of existing health services; 3. Knowledge & practices about disease(s) proposed to be screened; 4. Prevalence of important diseases with special reference to the diseases proposed to be screened; 5. Expected load of population likely to come up for screening, in respect of the community to be screened. 6. Resource analysis : Available medical and paramedical personnel, buildings, vehicles, equipment (for screening and final diagnosis), etc. What additional resources in terms of men, money and material will be required to smoothly undertake the screening test (and the final diagnostic test for those who are positive on screening)? 7. Are adequate treatment facilities available? 8. Reports on previous screening programmes which were undertaken in the same community earlier and the “weak areas” noticed. whether it is worthwhile and feasible to screen for the disease(s) in question: These considerations are very important before launching a screening programme. The important questions that you must ask yourself at this stage are: 1. Is the disease proposed to be screened an important health problem? 2. What are the high risk groups for the disease? 3. What is the prevalence in these groups? 4. Is a screening test available and can be administered to the subjects at a place near their home (say within 5 kilometres)? 5. Will the screening test be acceptable to the clientele? 6. Have the “diagnostic characteristics” (sensitivity, specificity etc.) of the screening test been worked out authentically? 7. Is a confirmatory test available? Will you be able to administer it to all those who are positive on screening test? 8. If the confirmatory test is to be given in a specialized centre, will that centre entertain your referred subjects? 9. Will the subjects be able to afford the travel and stay at the place of final diagnostic test? 10. Are you sure that there is a proper, proved modality of treatment for the disease you are going to screen? 11. Will those finally diagnosed be able to “afford” this treatment? Or will you be able to provide treatment out of governmental funds? 12. Are you sure that the disease you are screening for does not carry an over - riding “labelling” effect? 16 Identify the high risk groups: As we know, the prevalence will be high in high risk groups and hence the PPV and yield will be high; e.g. for screening for cervical cancer, “women > 35 years from lower socioeconomic status” may be identified as the high risk groups. Collect your logistics together: Remember, not to start a screening programme, in anticipation of the resources - you may cut a sorry figure and cause adverse publicity. First get call your required personnel, equipment, reagents, and other logistics ready. “Standardize” your personnel, instruments and techniques: The only method of ensuring a high repeatability of screening test is to centrally train your observers/ technicians, pre-test and certify them, standardize your equipment and reagents, and establish quality control procedures. Ensure community participation: Remember, a good epidemiologist never takes her community for granted. Your finest screening camp may not draw even a few subjects, simply because community participation had not been ensured. Contact the community leaders, peer groups and other members of the community who may matter, right in the planning stage itself. Explain to them the importance of the disease to be screened, and the usefulness of screening and early treatment. Emphasise on them that you need their active participation. Take their opinion as regards how they would like to get the camp organized. Give proper publicity: Make sure that at least 2 to 3 rounds of wide publicity have been undertaken, with an additional round of publicity for the high risk groups. The last round should ideally be undertaken 2 to 3 days before the start of screening camp. Ensure that those living in the remote, cut - off areas are covered well with your publicity - they are usually the ones who will benefit most by your screening programmes but are generally missed out by publicity campaigns. Conduct the screening programme: Do not leave things to chance. Be there yourself at the site, or at least ensure that one of your senior subordinates is there to address the “unforeseen” problems. Evaluate the screening programme: Set up your “criteria of evaluation well in advance. Write down your evaluation report at an early date while things are still fresh in the mind this will serve as a good reference document for subsequent Screening program. 17 References 1. Fletcher RH, Feltcher SW, Wagner E. Clinical Epidemiology -Thee essentials. 3rd edition. Copyrightl1l 1996 Williams & Wilkins, Maryland USA 2. Wilson JMG, Jungner G. Principles and practice of screening for disease. World health organization; France 1968. Public health papers vol 34 3. Silva IS. Cancer epidemiology: Principles and methods. World health organization. International agency for research on cancer Lyon, France 1999 4. Detels R, Beaglehole R, Lansang M, Gulliford M. Oxford textbook of public health. Vol. 2. The methods of public health. Fifth edition, 2009. Oxford university press, Newyork UK 5. Bonita R, Beaglehole R, Kjellström T. Basic epidemiology. 2nd edition © World Health Organization 2006 6. Bhalwar R. Textbook of public health and community medicine. 1st edition, 2009:Departtment of community medicine, AFMC Pune, New Delhi 7. Principles of Epidemiology in Public Health Practice- An Introduction to Applied Epidemiology and Biostatistics. Third Edition. Centers for Disease Control and Prevention (CDC), Atlanta. 8. Park K. Park’s textbook of preventive and social medicine. 20th edition, 2009. Banarsidas bhanot publishers, Jabalpur, India 18